In clinical trials, the primary goal is to evaluate the efficacy or safety of a treatment within a specific patient population. However, traditional statistical methods often fall short when unexpected events occur, which can lead to ambiguity in result interpretation. This is where the estimand framework, as outlined in the ICH E9(R1) addendum, steps in to offer a systematic approach to defining treatment effects that ensures clarity and consistency throughout the trial process.

The Challenge of Traditional Approaches

Traditional statistical analyses typically lack clear guidelines for handling intercurrent events (ICEs), leading to potential biases or ambiguous results. For example, if a patient discontinues treatment due to side effects or switches therapies, traditional analyses may inadequately address these occurrences, which can significantly impact the treatment effect interpretation. This lack of clarity can lead to differing conclusions among stakeholders—such as regulators, clinicians, or patients—about the treatment’s actual benefit.

Defining the Estimand Framework

The estimand framework enhances clarity in clinical trial data analysis. It is a formal, operationalized expression of the clinical question of interest, constructed with the following attributes

1. Treatment: Specifies the intervention and its comparators, such as placebo or standard care.
   Example: Comparing a new immunotherapy drug to conventional chemotherapy in a cancer trial.
   
   
2. Population: Defines the patient group included in the analysis.
   Example: Patients with advanced-stage breast cancer who have received prior treatment.
   
   
3. Variable: Indicates the outcome or endpoint, like survival rates or symptom reduction.
   Example: Occurrence of major cardiovascular events in a heart disease study.
   
   
4. Intercurrent Events (ICEs): Includes any events that might affect outcomes, such as treatment discontinuation, rescue medication use or death.
   Example: Accounting for rescue medication in a diabetes study.
   
   
5. Population-level Summary: Describes how outcomes between treatment groups are compared.
   Example: The summary measure could be a hazard ratio comparing the time to death between treatment and control groups in a survival analysis.

This structured approach ensures that all trial stakeholders have a common understanding of how results are derived, promoting more reliable and interpretable outcomes.

Estimands Strategies

A central aspect of the estimand framework is the targeted focus on possible ICEs and how they are dealt with. The estimand framework identifies five strategies to effectively address ICEs:

• Treatment Policy Strategy: This strategy includes all patient data for analysis, regardless of any intercurrent events that may have occurred.
• Hypothetical Strategy: Evaluates the treatment effect in a hypothetical scenario in which the observed ICE would not have occurred. It excludes data observed after ICEs and employs statistical methods to estimate the potential outcome in an uninterrupted scenario.
• Composite Strategy: Considers ICEs as integral parts of the endpoint definition, resulting in a composite endpoint that includes both the intended outcomes and the intercurrent events.
• While on Treatment Strategy: Focuses exclusively on data collected during the period when patients are actively receiving the treatment, disregarding any data after treatment discontinuation.
• Principal Stratum Strategy: Concentrates on specific patient subgroups defined by their response to treatment.

Application in Trial Design

Implementing the estimand framework

When using the estimand framework in clinical trials, careful planning is essential:

• Objective Alignment: Ensure the trial's goals align with the chosen estimand to address the right clinical questions.
• Handling ICEs: Predefine strategies for addressing events like treatment discontinuation or non-compliance.
• Estimation Methods: Specify statistical methods for an accurate treatment effect estimation.
• Sensitivity Analyses: Conduct analyses to test the robustness of results under different assumptions.

Benefits of Using Estimands

Using the estimand framework in clinical trials offers significant benefits by enhancing transparency and precision in defining trial objectives, improving the handling of intercurrent events, and aligning with regulatory expectations. It allows for the creation of analysis methods tailored to specific research questions, leading to a more realistic application of trial outcomes in real-world settings. Additionally, it reduces bias and provides a more accurate reflection of the true treatment effect, ultimately resulting in more robust, reliable, and meaningful trial results that better inform clinical practice and regulatory decisions.

Current Challenges

Implementing the estimand framework comes with several challenges as it is a relatively new concept. There is a need for ongoing education and training for statisticians, clinicians and other stakeholders to ensure a consistent and correct understating of the estimand framework. Communicating the estimand and its implications to non-statistical stakeholders requires careful consideration to ensure the concept is understood and applied correctly. Developing a protocol that clearly defines the estimand along with the strategies for design with intercurrent events can be time consuming and requires more detailed planning than traditional approaches.

Conclusion

The estimand framework offers a robust solution to the challenges of traditional clinical trial analysis by providing a clear, structured approach for defining treatment effects, especially in the presence of intercurrent events. By aligning trial objectives with regulatory expectations and enhancing transparency in data handling, the framework ensures that trial outcomes are both clear and credible, facilitating better clinical decision-making.

Our Experts Bio

Haiyin Wu, M.S.
Statistician, Amarex

Haiyin is a statistician with seven years of experience working with clinical data. She has developed extensive expertise across a wide range of study types and statistical methodologies. Her key strengths include sample size determination, statistical analysis planning, and the application of advanced techniques such as adaptive design, interim analysis, and simulation, all of which contribute to the strategic design and analysis of clinical trials. Actively involved in protocol design and development, she provides critical statistical input and contributes to the writing of trial protocols and research proposals. With strong programming skills and a deep understanding of statistical models for various data types, she is adept at customizing analytical approaches to meet the specific needs of each study.

References

ICH E9 (R1) addendum on estimands and sensitivity analysis in clinical trials to the guideline on statistical principles for clinical trials. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH). Updated Nov 20 2019.

Pohl, M., Baumann, L., Behnisch, R., Kirchner, M., Krisam, J., & Sander, A. (2021). Estimands—A Basic Element for Clinical Trials: Part 29 of a Series on Evaluation of Scientific Publications. Deutsches Ärzteblatt International, 118(51-52), 883.